Given this bit of free-form code

IDENTIFICATION DIVISION.
PROGRAM-ID. HELLO.

DATA DIVISION.
WORKING-STORAGE SECTION.
01 WS-TABLE.
05 WS-A PIC A(10)
OCCURS 1 TO 5 TIMES
DEPENDING ON D1
INDEXED BY I1.
01 D1 USAGE BINARY-CHAR UNSIGNED.

PROCEDURE DIVISION.
initialize ws-table.
set i1 to 10.
move i1 to d1.
move "thing" to WS-A(i1).
DISPLAY "ONE-D TABLE : "WS-TABLE.
STOP RUN.

This runs without error. Is that what's supposed to happen?

The GnuCOBOL version is

cobc (GnuCOBOL) 3.2-dev.0
Copyright (C) 2022 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <https://gnu.org/licenses/gpl.html>
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
Written by Keisuke Nishida, Roger While, Ron Norman, Simon Sobisch, Edward Hart
Built Jul 07 2022 09:17:27
Packaged Jul 07 2022 08:40:44 UTC
C version (MinGW) "9.2.0"

-Bruce

In article <c34be416-147f-44c4-a219-652768121e79n@googlegroups.com>,
Bruce Axtens <bruce.axtens@gmail.com> wrote:
>Given this bit of free-form code
>
>IDENTIFICATION DIVISION.
>PROGRAM-ID. HELLO.
>
>DATA DIVISION.
>WORKING-STORAGE SECTION.
>01 WS-TABLE.
> 05 WS-A PIC A(10)
> OCCURS 1 TO 5 TIMES
> DEPENDING ON D1
> INDEXED BY I1.
>01 D1 USAGE BINARY-CHAR UNSIGNED.
>
>PROCEDURE DIVISION.
> initialize ws-table.
> set i1 to 10.
> move i1 to d1.
> move "thing" to WS-A(i1).
> DISPLAY "ONE-D TABLE : "WS-TABLE.
> STOP RUN.
>
>This runs without error. Is that what's supposed to happen?

The table may be full of garbage with a thing in the last entry. Maybe
post the results you get, now, and what happens if you SET i1 TO 10 before
you INITIALIZE.

DD

Hello Bruce!

Wednesday June 21 2023 09:12, Bruce Axtens wrote to All:

> Given this bit of free-form code

> IDENTIFICATION DIVISION.
> PROGRAM-ID. HELLO.

> DATA DIVISION.
> WORKING-STORAGE SECTION.
> 01 WS-TABLE.
> 05 WS-A PIC A(10)
> OCCURS 1 TO 5 TIMES
> DEPENDING ON D1
> INDEXED BY I1.
> 01 D1 USAGE BINARY-CHAR UNSIGNED.

> PROCEDURE DIVISION.
> initialize ws-table.
> set i1 to 10.
> move i1 to d1.
> move "thing" to WS-A(i1).
> DISPLAY "ONE-D TABLE : "WS-TABLE.
> STOP RUN.

> This runs without error. Is that what's supposed to happen?

> The GnuCOBOL version is

> cobc (GnuCOBOL) 3.2-dev.0
> Copyright (C) 2022 Free Software Foundation, Inc.
> License GPLv3+: GNU GPL version 3 or later
> <https://gnu.org/licenses/gpl.html> This is free software; see the
> source for copying conditions. There is NO warranty; not even for
> MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. Written by
> Keisuke Nishida, Roger While, Ron Norman, Simon Sobisch, Edward
> Hart Built Jul 07 2022 09:17:27 Packaged Jul 07 2022 08:40:44 UTC
> C version (MinGW) "9.2.0"

Why are you expecting it to produce an error ?

Although the value used is outside the table range unless you have full
diagnostic checks set on compile it would not produce a error condition but
doing so produces :

cobc -x test1.cbl -d -g
-bash-5.1$ ./test1
libcob: test1.cbl:15: error: OCCURS DEPENDING ON 'D1' out of bounds: 0
note: minimum subscript for 'WS-A': 1

Last statement of "HELLO" was INITIALIZE at line 15 of test1.cbl
ENTRY HELLO at test1.cbl:15
Started by ./test1

I only added 1st line with " >>source free"
before compiling.

Without the runtine diags it may well show garbage data but here did not
as showed :

-bash-5.1$ cobc -x test1.cbl
-bash-5.1$ ./test1
ONE-D TABLE :
thing

version used :

cobc -V
cobc (GnuCOBOL) 3.2-dev.0
Built May 30 2023 21:15:38
Packaged May 30 2023 16:41:34 UTC
C version "10.4.0"

Vincent

On Thursday, 22 June 2023 at 2:14:35 am UTC+8, Vincent Coen wrote:
> Why are you expecting it to produce an error ?
I suppose because many other languages do. I was not aware of the compiler switch and didn't find it after a couple of searches (maybe I didn't search for the right keyword).

The fact that it doesn't produce an error means I just have to be more careful (or find the relevant switch).

-Bruce

On Friday, June 23, 2023 at 9:29:21 PM UTC-5, Bruce Axtens wrote:
> On Thursday, 22 June 2023 at 2:14:35 am UTC+8, Vincent Coen wrote:
> > Why are you expecting it to produce an error ?
> I suppose because many other languages do. I was not aware of the compiler switch and didn't find it after a couple of searches (maybe I didn't search for the right keyword).
>
> The fact that it doesn't produce an error means I just have to be more careful (or find the relevant switch).
>
> -Bruce
Up until version 6.1, the IBM COBOL compiler and runtime environment were very sloppy in allowing this sort of thing. Then in 6.1 when they put in checks for such things, all kinds of issues were found in existing code. The migration to 6.1 is quite a big deal for many shops.

In article <e780a289-5a70-4277-9c71-374b6e44ee3an@googlegroups.com>,
William Fahle <billfahle@gmail.com> wrote:
>On Friday, June 23, 2023 at 9:29:21???PM UTC-5, Bruce Axtens wrote:
>> On Thursday, 22 June 2023 at 2:14:35???am UTC+8, Vincent Coen wrote:
>> > Why are you expecting it to produce an error ?
>> I suppose because many other languages do. I was not aware of the
>compiler switch and didn't find it after a couple of searches (maybe I
>didn't search for the right keyword).
>>
>> The fact that it doesn't produce an error means I just have to be more
>careful (or find the relevant switch).
>>
>> -Bruce
>Up until version 6.1, the IBM COBOL compiler and runtime environment
>were very sloppy in allowing this sort of thing. Then in 6.1 when they
>put in checks for such things, all kinds of issues were found in
>existing code. The migration to 6.1 is quite a big deal for many shops.

Before there was enough core to allow Big Tables in WORKING-STORAGE some
people would define smaller tables consecutively. It wasn't a difficulty
as long as the Senior Programmer made sure youngsters with less than five
years' experience didn't deal with that code... or until the shop
installed one of those fancy, new compilers that'd re-arrange storage
areas.

Ahhhh, for the Oldene Dayse... did'ja hear that core just fell below the
buck-a-byte level?

DD

Hello William!

Thursday July 06 2023 19:49, William Fahle wrote to All:

> On Friday, June 23, 2023 at 9:29:21 PM UTC-5, Bruce Axtens wrote:
>> On Thursday, 22 June 2023 at 2:14:35 am UTC+8, Vincent Coen
>> wrote:
>> > Why are you expecting it to produce an error ?
>> I suppose because many other languages do. I was not aware of the
>> compiler switch and didn't find it after a couple of searches (maybe
>> I didn't search for the right keyword). The fact that it doesn't
>> produce an error means I just have to be more careful (or find the
>> relevant switch). -Bruce
> Up until version 6.1, the IBM COBOL compiler and runtime environment
> were very sloppy in allowing this sort of thing. Then in 6.1 when they
> put in checks for such things, all kinds of issues were found in
> existing code. The migration to 6.1 is quite a big deal for many
> shops.

IBM compilers have always had a diag check facility on way or another BUT
it does depend on what settings are preset when the compiler was built for
the site. As you no doubt know there are very many that can be so the
sysprog that sets the new compilers up does so according to a site wide
specification created by, hmm well someone any way :)

I use 3 or 4 of them and they all have different settings pre-configured
in.

GnuCobol is not so easy to do and I have requested a more automated way of
doing so other than the compiler config files via std=<dialect> etc.

For more settings run cobc -h

look for -d (or --debug) -g for enable C compiler debug and stack check

For tracing the program add -ftraceall when environment var COB_SET_TRACE
is set to 1 or Y.

For more details read the Programmers Guide or Reference manuals.

Vincent

In article <1688692562@f1.n250.z2.fidonet.ftn>,
Vincent Coen <VBCoen@gmail.com> wrote:
>Hello William!
>
>Thursday July 06 2023 19:49, William Fahle wrote to All:
>
> > On Friday, June 23, 2023 at 9:29:21???PM UTC-5, Bruce Axtens wrote:
> >> On Thursday, 22 June 2023 at 2:14:35???am UTC+8, Vincent Coen
> >> wrote:
> >> > Why are you expecting it to produce an error ?
> >> I suppose because many other languages do. I was not aware of the
> >> compiler switch and didn't find it after a couple of searches (maybe
> >> I didn't search for the right keyword). The fact that it doesn't
> >> produce an error means I just have to be more careful (or find the
> >> relevant switch). -Bruce
> > Up until version 6.1, the IBM COBOL compiler and runtime environment
> > were very sloppy in allowing this sort of thing. Then in 6.1 when they
> > put in checks for such things, all kinds of issues were found in
> > existing code. The migration to 6.1 is quite a big deal for many
> > shops.
>
>IBM compilers have always had a diag check facility on way or another BUT
>it does depend on what settings are preset when the compiler was built for
>the site.

SSRANGE would check for SubScriptRANGE. The advantage was that it avoided
table overflows, the disadvantage was that it chewed up more resources
during compiles.

https://stackoverflow.com/questions/56866908/gnucobol-equivalent-of-nossrange

DD